Abstract
This study explores the thematic content of news headlines and assesses their predictive power for corporate bond defaults. It establishes a data-driven framework, that emphasizes transparency and interpretability through the incorporation of explainable AI (xAI) techniques. The interpretable AI method, BERTopic, is applied to analyze news headlines from Jan. 1, 2014, to Aug. 31, 2021. A total of 18 economically inrerpretable topic measures are derived by combining similar topics among 78 original topics, offering insights into bond issuers’ operational behavior and associated risks. Integrating the BERTopic model, various machine learning prediction models, the SHapley Additive exPlanations (SHAP) approach, and feature combination approaches, this study uncovers the incremental information contributed by news headlines beyond financial ratios and economic variables. The inclusion of these topic measures significantly enhances the predictive performance of first-time corporate bond defaults within a 3-month horizon. Additionally, the robustness of news headlines’ information value is validated by extending the sample and employing an alternative study sample with differing credit risk scenarios, diverse markets, and even distinct news sources.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
